Health and Education: Another Look with the Proper Data

During the XXth century, life expectancy levels have converged across the world. Yet, macroeconomic studies, as Acemoglu and Johnson (2007), estimate that improvements in health have no impact on growth or any factors of growth; in particular, they find no impact of life expectancy increases on education. We argue that their pessimistic results with respect to schooling investment is due to the use of an imprecise proxy. Indeed, when life expectancy increases at time t, only the cohort born at t should increase its human capital investment. On the contrary, Acemoglu and Johnson (2007) look at the impact of life expectancy improvements on the average education of the whole population aged above 15, which evolves much slower. We have reproduced their estimations with a cohort-based measure of education and find a positive and significant impact of life expectancies on education, of a magnitude between 20% and 47%. Finally, we use both the Cohen-Soto (2007) and the Barro-Lee (2010-2013) databases on education, and explain in the text why the former delivers better results than the latter.